Category: projects

Civil, Environmental and Architectural Engineering of the University of Padua We are currently using CloudVeneto to run Finite Element simulations on cracking due to drying using the poromechanical and the phase-field formulation. Basically we use Ubuntu system and install the deal.II open source software. The flexible structure of the Cloud allows us to create ourselves virtual machines and install the open source libraries we need. The access is straightforward and the support is very helpful when help is needed. Currently we are using the largest vm to run some tests on desiccation using the phase-field approach. . PhD candidate Tuanny Cajuhi and Prof. Laura de Lorensis (University of Technology in Braunschweig) and Prof. Lorenzo Sanavia    

Department of Biomedical Sciences of the University of Padua This project is part of an international competition (CAFA). The input dataset consists of 130,000 unknown protein sequences. The goal is to predict the functional characteristics of each of these sequences as accurately as possible. It is well established that proteins with similar amino acid sequences share molecular structure and therefore biological function. The strategy employed involves searching for sequences similar to those that are unknown in databases and subsequently transferring annotation. This is done using software for local alignment optimized for rapid searching in databases, and functional annotation is transferred based on sequence similarity. The CloudVeneto platform has allowed for the optimization of this search process by parallelizing the workload across available resources.   Dr. Ivan Mičetić

Department of Chemical Sciences of the University of Padua During last decades nanotechnology grew widely on several fields from energy to medicine. Nowadays, the so called nanobiotechnology and nanomedicine are opening new horizons on diseases monitoring and personalized therapies. Scaling down noble metals into nanometric structures produces interesting and, sometimes, unexpected properties that are not effective on the macrometric scale. The case of gold nanostructure is one of the most studied because of their characteristic chemical inertia that prevents undesirable side effects into the hosting organism. At the same time, these nanostructures can be tuned to interact strongly with infrared light, that is considered a safe and useful source for diagnosis and drugs-free therapies. The advent of high performing computational tools completely revolutionized the way in which such nanostructures are designed and produced. Simulating the interaction between complex nanostructures between each other and light helps the synthesis of materials with optimized performances. Our research projects are involved in computationally designing and then producing gold nanostructure as efficient sensors for diagnosis and for drug-free therapy. We are using the CloudVeneto resources for such use case   Prof. Moreno Meneghetti

Department of Physics and Astronomy of the University of Padua We have used Fortran programs, developed by us, to calculate tables (databases) of coefficients to be used in calculations of many-body systems physics, particularly for nuclear systems composed of few nucleons or cluster nuclei. For now, we have limited ourselves to a 1D model, obtaining benchmark results that are being prepared for publication, and the calculations have been relatively inexpensive. However, with the refinement of the 3D version of the code, we will have a powerful tool for nuclear structure calculations and to verify numerous theoretical hypotheses. It will also be possible to extend the scope of application to atomic and molecular systems, expanding the interdisciplinary nature of the project. The use of the cloud has proven useful for distributing the calculations. Prof. Lorenzo Fortunato

Civil, Environmental and Architectural Engineering of the University of Padua Within the framework of the European project ITN2015 INFRASTAR, Geotechlab utilizes CloudVeneto to perform particle simulations of granular soils. Specifically, it leverages the available resources to conduct batch triaxial cyclic tests on virtual soil samples to model ratcheting phenomena, i.e., the accumulation of irreversible deformations. These models aim to understand how the mechanical behavior of the soil changes after millions of cycles of loading and unloading, and find application in the long-term design of structures such as wind turbine foundations or road infrastructure foundations. The capabilities of this cloud include the allocation of large computing resources tailored to the computational needs of the model, the ability to launch multiple virtual machines remotely, and the availability of large volumes to allocate the output files of the simulations. PhD candidate Gianluca Zorzi and Dr. Fabio Gabrieli

Department of Physics and Astronomy of the University of Padua The Cosmology group of the Department of Physics and Astronomy uses CloudVeneto to numerically solve integrals and differential equations necessary for studying the temporal evolution of various cosmological observables, including anisotropies and spectral distortions of the cosmic microwave background radiation. Studying these observables has allowed us to produce forecasts regarding the capabilities of various future experiments (e.g., the PIXIE and CORE satellites) to provide information on various inflationary models. We particularly appreciated the platform’s ability to create virtual machines with various numbers of cores and varying amounts of RAM. Given the large amount of memory required by our specific applications, they are difficult to optimize on cluster systems. The ability to use many cores with shared memory has allowed us to fully utilize the available resources. Equally appreciated is the ability to autonomously manage our virtual machines. Dr. Andrea Ravenni

Civil, Environmental and Architectural Engineering of the University of Padua The research group ‘Hydrological Data Assimilation’ utilizes data assimilation (DA) techniques, whose main purpose is to directly integrate information from experimental observations into hydrological models, thus obtaining an optimal estimate of the system state and parameters. DA techniques enable updating model predictions with measured data as soon as they become available, also providing an estimation of errors and uncertainties in simulations. Among the various DA techniques available, the Ensemble Kalman Filter (EnKF) is of particular interest, thanks to its ability to handle nonlinear problems through a Monte Carlo approach, which involves approximating the statistical distributions of input/output data using hundreds or thousands of parallel simulations. For this reason, EnKF typically requires significant computational resources, both in terms of data storage and the number of CPUs to be used. The CloudVeneto infrastructure allows us to fully exploit DA techniques and develop cutting-edge scientific tools for solving complex hydrology problems. Dr. Matteo Camporese

Department of Information Engineering of the University of Padua The IMS group of the Department of Information Engineering uses the CloudVeneto platform to conduct information retrieval experiments. In particular, automatic indexing of large collections of textual documents involving high-frequency input/output operations is performed. The created indexes are maintained in secondary memory and used to perform retrieval operations using various search engines (several hundred). Additionally, for each retrieval operation, we calculate dozens of evaluation measures (such as average precision) on which we perform statistical analyses. The large amounts of processed data require the availability of a large amount of main memory and secondary memory. We have used Java and Matlab libraries. Dr. Gianmaria Silvello